Two competing models for the role of water vapor feedback in global warming have been proposed : LindzenÕs cumulus drying hypothesis which results in a strong negative feedback from Climatically invariant (or fixed) relative humidity assumption proposed, originally by Arrhenius and Manabe/Wetherald, which results in a moderately strong positive feedback. The drastic difference in the predicted feedback effect should have made it possible to test the validity of the two models but a test has not been attempted yet. A suitable natural experiment, originally suggested by Lindzen to test general circulation models, is the annual cycle of surface temperature (Ts) for which the global mean amplitude is as large as 4K. Corresponding changes in the annual cycle of Ts, vertical water vapor distribution, w and atmospheric greenhouse effect Ga should provide an important test. Towards this goal, this study employs a multitude of global data sets (satellite, surface and in-situ) for Ts, vertical water vapor distribution and atmospheric greenhouse effect Ga,that have been made available recently. We adopt the radiometric definition for Ga (difference between surface emission and outgoing long wave radiation) since it is the only data set that includes the effects of upper tropospheric water vapor to the accuracy that we need for this test.
We focus our attention on the tropics, since the interactions between deep convection and Ts play a central role in both the theories. The new data encompass a global domain including both the continents and oceans, as well as the ascending and descending branches of the Walker and Hadley cells. We show that Ga, w and Ts variations during the annual cycle are highly correlated. The feedback sensitivity factor, d Ga/dTs, tracks very well the response of equatorial convection and large scale circulation of water vapor. The main features of our results are: (i) Increase in equatorial convection in response to the surface warming moistens the mid-troposphere between 10N to 10S and contributes to the so-called super greenhouse effect; the role of large scale dynamics in this moistening requires further study. ii) The Hadley cell response to the equatorial convection is to dry the sub tropics, which compensates for some of the enhanced equatorial greenhouse effect; and iii) On tropical scales (including the equatorial and sub-tropical regions) however, there is a net moistening effect and water vapor exerts a positive feedback effect, somewhat similar to that of the fixed relative humidity models. However, the annual cycle change in Ts are accompanied by tropical lapse rate changes unlike those simulated by GCMs. With respect to the cumulus drying model, the data presented here do not support the large scale (tropical) negative feedback effect.
The new data suggest that regulation of Ts in the tropics should be considered on two scales: First is the convectively active warm low latitude oceans where the super greenhouse effect necessitates a thermostat type regulation; and second is the tropical mean scale, where the standard radiative-convective model has to be extended to account for transport to the extra-tropics. The important role of tropical convective-cirrus clouds, in particular the role of excess solar absorption, on tropical mean climate is most likely to regulate the mean tropical surface evaporation.